Binomial logisting regression with 2 level 2 variables

[FatimaGhani]

Hi there,

I'm trying to run the following model, but I get an error message in mlwing and in stata. I'm getting the following message: "The model did not run properly in MLwiN. Re-run the model without the nopause option to debug the model in MLwiN". Also, I was wondering if I need to have MeanPerneighcohe as level 1 in the model? The corresponding level 1 variable is Perneighcohecat (3 categories).

Many thanks,

**** MODEL 3 - M2 + nd_5 + newedcat + newincome + newocc

quietly runmlwin aawlktotguide cons sex age45to49 age50to54 age55to59 age60to65 nd_richer nd_middle nd_poorer nd_poorest neweduDip neweduCer neweduNone newoccWhite newoccBlue newoccHome newoccReti newoccMiss newincome2 newincome3 newincome4 newincome5 newincomeMiss, level2(cd2006: cons MeanPerneighcohe) level1(id:) discrete(distribution(binomial) link(logit) denom(cons) pql2) nopause
runmlwin aawlktotguide cons sex age45to49 age50to54 age55to59 age60to65 nd_richer nd_middle nd_poorer nd_poorest neweduDip neweduCer neweduNone newoccWhite newoccBlue newoccHome newoccReti newoccMiss newincome2 newincome3 newincome4 newincome5 newincomeMiss, level2(cd2006: cons MeanPerneighcohe) level1(id:) discrete(distribution(binomial) link(logit) denom(cons)) mcmc(orthogonal) initsprevious nopause nogroup
runmlwin, noheader noretable or
estimates store Model3

[ChrisCharlton]

In your example you are running two models (one with IGLS and one with MCMC). Which of these are you getting the error with? If the IGLS model runs without error can you post the output (without the quietly prefix) for this model? If this model doesn't run either then I would suggest simplifying your model until you find one that works, which might indicate which variables are being problematic.

MLwiN does not allow the specification of random variables at level one for discrete models.